FR2512293A1 - METHOD FOR ENCODING FREQUENCY AND DEVICE GENERATING TONES OBTAINED BY THE METHOD - Google Patents

Abstract

THE PRESENT INVENTION RELATES TO THE DIGITAL TYPE FREQUENCY GENERATION, THE SAID FREQUENCIES BEING ABLE TO TAKE VALUES IN THE BAND 0, 4000HZ. THE INVENTION ESSENTIALLY CONSISTS OF USING THE PERIODICITY OF THE SINE FUNCTION ON (0, 2P), IN MAKING EACH FREQUENCY F CORRESPONDENCE OF THE SAMPLES, EACH SAMPLE BEING REPRESENTED BY A TRIPLET P, Q, S, P REPRESENTING THE RANK, Q BECAUSE OF THE QUADRANT TO WHICH THE SAMPLE BELONGS TO (0,2P) S THE SIGN, TO MEMORIZE 14RT SAMPLES OF THE 1HZ SINUSOID ON (0, P2), T BEING THE SAMPLING PERIOD, R BEING AN ENTIRE NATURAL SINUSOID 1 HZ ON (0, P2), T BEING THE SAMPLING PERIOD, R BEING A NATURAL SINUSOID 8000 DIVIDER OF 1 TO MATCH ANY Y SAMPLE REPRESENTATIVE OF THE FREQUENCY F (YSIN 2 P PT) A TRIPLET P, Q, S SAMPLE CHOSEN FROM THE 14RT SAMPLES, THE SAID CORRESPONDENCE BEING TWO UNIVERSAL. APPLICATION TO THE GENERATION OF TONES, TO INTERCENTRAL SIGNALING, ETC.

Description

The present invention relates to the generation of frequency

  digital type, said frequencies being able to take values in the band I 0, 4000 Hzl

  Such frequencies apply in particular to the

  Multi-frequency signals, tones, etc. The tones, in effect, result from clocking frequencies while the multi-frequency signals, for example those conforming to the SOCOTEL, R 2, keyboard codes, result from the combination of

various frequencies.

  The signal to be generated is a signal of general form E = sin 2 Ti ft where f is an integer having the dimension

  frequency in PCM systems (pulsed modulation)

  and encoding), any signal characterized by a frequency f is represented by a series of numerical values (xn) nnn called samples Each sample xn is written as: Xn = sin 2 WT fn T o N is the rank of the sample and T is the period of the sampling One usually chooses in the systems MIC 1 / T = 8000 Hz, which limits f to 4000 Hz

(Shannon's theorem).

  A multifrequency signal generator device has already been described in the review Switching and Electronics No. 59 of October 1977 pages 99 to 115 This generator also aims to use the frequency of the sine function on the interval (0.2 W) in representatives this interval) M values sin o, = 2 W Ti defined by 25 defined by M, numbered values and entered in read-only memory This makes it possible to store only M '= values corresponding to the interval (O, L) If we read at the frequency F the samples of the memory, we obtain a

  representative samples of a sinusoid of frequency

F

  However, the number M in this document is a

of 2.

  M = 2 A reading of the table all k values, k being a real number, provides a sequence of samples representing a sinusoid of frequency fk F with O Mk <is still 0 <% kn (1 M 2 0 <k < 2 n-1

  However, the M values of k are not values

  samples and read, only the whole

  binary elements is taken into account so that we introduce

  This method causes a misunderstanding

  to define the representative sample sequence -

the desired frequency.

  The object of the present invention is to provide a tone generator which establishes a one-to-one correspondence between the samples representative of the 1 Hz frequency defined on (0, 27) and to express any signal of full frequency f defined on

  , 4000 Hz, from the 1 Hz signal.

  The invention uses the knowledge of the samples Y de-

  n finite for f = 1 Hz in the invert O n, 8000 to determine the set of numerical values of all the suites (Zn) _ +

  for frequencies f such as 1 i C 4000.

  For this purpose, according to the invention, a number R of

  samples such that R is related to the frequency f = T of samples

1 T

  such that f = T = 4 r R, where r is a natural integer divided by

  The most general case is to choose T

  r = 1 which requires R 2000 samples.

  The method according to the invention essentially consists of storing 4 samples of the 1 Hz sinusoid on (0,) 4 r T '2 T being the sampling period, r being a natural integer

  divisor of T = 8000 and to match any sample

  YN representative of the frequency f to generate (YN = sin 2 n PT),

  a triplet sample (P, Q, S) selected from the 4 r T -chan-

  the correspondence is one-to-one.

  The device according to the invention comprises means for storing 4 samples of the sinusoid 1 Hz on (0, e), where T 1 being a natural integer, means for matching any representative sample of each frequency forming a tone, two samples of rank Pl and P 2, represented by the triplets (Pl, Qi '51) and (P 2 Q 2' 52) among the 1 samples, VQ 28 '52 4 r T said correspondences being one-to-one, T being the period

  sampling, means to characterize the durations of

  mission or non-emission by another dynamic parameter D, means for deducing from the triplet (P1, Qi '51) (resp (P 2, Q 2, 52) of the preceding state and of the frequency f 1 (resp 2) forming the tone, the new triplet (P'1, Q'i 'S'I) (resp (P'2 Q'2, S'2)

  characteristic of the frequency f 1 (resp f 2).

  The invention makes it possible to code each signal without introducing

error.

  Other features and benefits will appear at the

  reading the following description illustrated by drawings.

  Figure 1 shows the properties of the sine function in the interval (0, 2 T) allowing the definition of any frequency f e (1, 4000 Hz).

  FIG. 2 represents a diagram of the device of general

  tones according to the invention.

  Figure 3 shows the flowchart of the calculation of the triplet

  (P 'Q' S ') from the triplet (P, Q, S).

  Referring to Figure 1 on the trigonometric circle, we can represent in the four quadrants the properties

of the sine function.

  For each quadrant we can associate a number K such that O <K <3 The samples YN placed in each of the quadrants are such that: YN + 8000 k = YN We can easily match to any sample of rank N another sample of rank P For example in

the first quadrant K = O -

  0 N 8000 k <2000 YN = sin 2 i TNT = sin 2 ITP T with: N = P + 8000 koi P <2000 Similarly for the other quadrants, for K = 1 2000 N 8000 k <4000 YN = sin 2 r NT = sin ('2 IPT) = sin 2 TPT YN = sin 2 Tr (4000 P) T therefore: YN = sin 2 ITP TN = 4000 P + 8000 kb P 2000 And again for the third quadrant K = 2 and N 000 N80 k 6000 the YN = sin 2 TN T = sin (Tr + 2 IT PT) = sin 21 TPT = sin 2 IT (4000 + P) T therefore: I YNI = sin 2 'TF PT with N = 4000 + P + 8,000 k

O AP _ 2000

  Finally in the same way for the fourth quadrant: IYNI = sin 2 T-T P with N = 8000 P + 8000 k

0 P <& 2000

  Thus any sample YN can be characterized on the one hand by the quadrant to which it belongs, quadrant defined by a number K (OK, 3), on the other hand by a number P-which makes it possible to determine the absolute value of the sample. and such that: YNI = sin 2 R PT with OP <2000 It is therefore appropriate to code any integer frequency f between 1 Hz and 8000 Hz by three parameters S, Q, P, where S represents the

  sign of the sample stored in memory, Q its parity and P its rank.

  The parameters (Q, S, P) will constitute the dynamic definition

  of the sample YN Any new frequency to be generated

  corresponds to another triplet (P ', Q', S ') A calibration algorithm

  ass defines the new triplet (Q ', P', S ') from the frequency f to be generated and from the previous state (P, Q, S)

  as will be explained later.

  To the sample of rank N of the signal E = Am sin 2 Trf t, one can associate the sample YN defined by the triplet P, Q, S such that En = Am YN o Am is the maximum amplitude exploitable by the system chosen appropriately: J En} = Am sin 21 TPT

  FIG. 2 represents an embodiment of a general

  tone transmitter implementing the method of the invention.

  The samples of the frequency 1 Hz are put in a memory 1 of capacity 4 r T o is the sampling frequency, r is

  a natural integer divisor of either 8000 is chosen advantageously

  This choice gives the maximum precision in the definition of the frequencies.

  memory since the sampling frequency is 8 kHz.

  In practice, 2048 words cxoe memory dimension is advantageously chosen, which simplifies the calculation algorithm of the

new triplet t P 'Q' S ').

  The frequency coding method according to the invention makes it possible in particular to generate tones. It is well known that a tone TON is expressed as a function of time as a sum of signals of frequencies f 1 and f 2 TON = LA 1 sin 2 1 Tf 1 t + A 2 sin 2 T f 2 t 3 h (t) o A, and A 2 are the respective amplitudes of the frequencies fl and f 2 and h (t) is a function of period (t 1 + t 2 ) such that h (t) = 1 over Ot 1) h (t) = O over (t 1, t 2 + t 1)

  t 2 can be zero This function h (t) corresponds to the cadence-

  certain telephone tones.

  In the same way as for a single frequency, at each

  frequency f 1 or f 2 corresponds to a triplet of dynamic parameters

  ques (Pi, Qi '51) or (P 2 Q 2' 52) which are put in a memory

live 2.

  The various parameters A 1, A 2, wire f 2 't 1, t 2 are put in a dead memory 3 Thus two new dynamic parameters D and C are associated with the signal in the case of a tone, D representing the transmission state or the locked state and C representing the current value of a counter measuring the duration of emission

(t 1) or non-emission (t 2).

  The device of the invention essentially comprises three memories, a memory 1 dead samples of the sinusoid sin 2 i Tt at 1 Hz, a memory 2 alive parameters P, Q, S, C, dynamic D of each frequency forming the tone and finally a read-only memory 3 of the characteristics A 1, A 2, f 2, t 1, t 2 of the same tone or of the frequency f to be generated. The memory 2 supplies at the input of the memory 1 an address representing the

rank P of sample YN-

  This sample YN from the memory 1 is applied in a register 4 which itself provides it at the input of a calculation unit 5 A register 6 receives the sign S of the samples as well as the characteristics of the tones making it possible to define the

  type of operation (pure frequency or dual frequency code to be

  kill) This result applied to the input of the calculation unit 5, defines the operation) addition or subtraction carried out by this unit 5 The information of the memory 1 of the samples of

  the sinusoid at 1 Hz are addressed by the rank P of the samples

  lons provided by memory 2 This YN sample from the

  moire 1 is applied to the input of the register 4 of the samples before being applied to the input of the calculation unit 5. This sample RECH contained in the register 4 must in fact undergo

  a number of offsets and additions equivalent to a multiplicity

  cation. The dynamic P parameter from memory 2 is also

  applied to the input of a register 9 memorizing the rank P of the

  previous sample The Q parameter is applied to

  the input it of a device for defining the operating sequence

  This parameter Q triggers the running of the algorithm, which will be explained in detail later:

  device it outputs an indication of a desired operation.

to the calculation unit 13.

  The device 10 also receives at its input the character

  the width of the hollow or the full, corresponding to the frequency

  quence f or at the frequencies f 1 f 2 of the tone concerned, resulting from the read-only memory 3 At the output of the timing device 10, the characteristic of recess or full is applied on the one hand,

  at the input of the multiplexer 12, on the other hand at the input of the device.

  It defines the sequence of operations.

  The output of the multiplexer 12 is the input of the memory

  ve 2 o are recorded the dynamic states P, Q, S, C, D of the signals to be generated 7

  As has already been mentioned, every frequency is gener-

  from the previous state The calculation algorithm makes it possible to define the new parameters P ', Q', S ', C', D 'from the old P, Q, S, C, D This algorithm is realized thanks to

various registers.

  The rank of the preceding sample set in the register 9 is applied to the input of a calculation unit 13 which performs

  the operation defined by the device 11 This device 13 re-

  also the frequency characteristic resulting from the

  memory 3 after registering in a register 14 At output the calculation unit 13 provides an intermediate result which is

  applied to the input of a multiplexer 15 which multiplexes this result.

  tat with the rank P of the previous sample from the memory 2 The register 9 thus memorizes not exactly the rank P of the previous sample from the memory 2 but the result of the

  multiplexing from the multiplexer 15, multiplexing linked to the routing

the calculation algorithm.

  The result of the calculating unit 13 which is the rank P 'of the

  new sample is also applied to the input of the multi-

plexeur 12.

  The parameters Q 'and S' respectively of quadrants and signs of the samples are provided at the output of the device.

  defining the sequence of operations.

  Thus the multiplexer 12 receives the new parameter P 'from 13, the parameters Q' and S 'from there and the parameters C' and D 'from 10 All these new parameters are applied to the input of the random access memory 2 On leaving the

  memory 2 the sign S is also applied to the input of the regis-

  6 definition of operations.

  An amplitude processing device 7 receives as input the frequency characteristics coming from the memory 3 and corrects the result stored in a register 8 containing the calculation result provided by the calculation unit. The intermediate results provided by this register 8 are reintroduced at the input of the computing unit 5 until the end of the operations defined by the device 6, thereby realizing, by a

  result of shifts and additions, multiplications.

  The final result is again subjected to a linear compression in logarithmic compression by means of a device 16 before being outputted in PCM code according to the specifications

inherent in this code.

  Referring to FIG. 3, the calculation making it possible

  of any sample YN 'knowing the preceding sample YN represented by the pair (K, P) K being the quadrant and P the

  rank, is illustrated by a flowchart.

  The new sample YN 'is defined by the pair (K', PI).

  We have previously seen that N '= N + f and 1 <f <4000 We will study the 4 possibilities differentiated by the value of K associated with the sample YN' So when K = YN is in the first quadrant, we

  have N = P so N '= N + f = P + f.

  Case where O <P + f <2000 YN 'is in the first quadrant so {K' = 0, P '= P + f and sin 2 lTN'T = sin 21 TP' -T Case b 2000 <P + f < 4000 y N 'is in the 2nd quadrant N' = 4000 P '= P + f so K' KI P '= 4000 (P + f) and sin 2 Il N' T = sin 2 Ir P 'T Casc 4000 < P + f <6000 YN is in the third quadrant N '= 4000 + P' = P + f so (K '= 2 P' = (P + f) 4000 (and sin 2-1 N 'T = sin 21 TP 'T When K = YN is in the 2nd quadrant We have YN

N = 4000 P

  N '= N + f = 4000 P + f Case at 2000 <4000 P + f <4000 YN' is in the 2nd quadrant N '= 4000 P' = 4000 P + f so (K '= 1 P' = P f isin 2 lT N 'T = sin 2 ITP' T Case b 4000 <4000 P + f $ 6000 YN 'is in the 3rd quadrant N' = 4000 + P '= 4000 P + f so (K' = 2, P '= f P lsin 2 ri N' T = sin 2 ITP 'T

12293

  Case c 6000 <4000 P + f 8000 YN, is in the 4th quadrant N '= 8000 P' = 4000 P + f so $ K '= 3 P' = 4000 + Pf (sin 2 IT N 'T = sin 2 i TP 'T Then when K = 2, YN is in the 3rd quadrant We have:

N = 4000 + P

  N '= N + f = 4000 + P + f Case at 4000 <4000 + P + f _ 6000 YN' is in the 3rd quadrant N '= 4000 + P' = 4000 + P + f so (K '= 2 , P '= P + f isin 2 1 NOT = sin 2 Tr P' T Case b 6000 <4000 + P + f <8000 YN 'is in the 4th quadrant N' = 8000 P '= 4000 + P + f (K '= 3 P' = 4000 (P + f) sin 2 1 TN 'T = sin 2 Tr P' T Case c 8000 <4000 + P + f <2000 + 8000 YN, is in the first quadrant N '= P' + k 8000 with k = 1 N '= P' + 8000 = 4000 + P + f therefore (K '= OP' = P + f 4000 lsin 2 1 TN'T = sin 2 -i P 'T For K = 3, YN is in the 4th quadrant We have:

N = 8000 P

  N '= N + f = 8000 P + f Case at 6000 <8000 P + f <, 8000 YN' is in the 4th quadrant N '= 8000 P' = 8000 P + f so K '= 3 P' = Pf (sin 2 TN'T = sin 27 WP 'T Case b 8000 <8000 P + f <2000 + 8000 YN, is in the first quadrant N' = P '+ 8000 with k = 1 N' = P '+ 8000 = 8000 P + f so (K '= 0, P' = f P isin 2 1 t NOT = sin 2 Tr P'T

25.12293

  Case c 2000 + 8000 1? 8000 P + f <4000 + 8000 YN is in the 2nd quadrant N '= 4000 P' + 8000 with k = 1 N '= 4000 P' + 8000 = 8000 P + f so (K '= 1, P' = 4000 + Pf isin 2 i N 'T = sin 2 IT P' T We can deduce from the cases exposed above that: Finally, for K = O and K = 2 The passage of the couple (K, P) to the pair K 'P ') follows the same law. casa: K '= K, P' = P + f case b: K '= K + I, P' = 4000 (P + f) case c: K '= K + 2, P' = (P + f) 4000 for K = 1 and K = 3 The transition from the torque (K, P) to the pair (K ', P') follows the same law. casa: K '= K, P' = Pf case b: K '= K + l, P' = fP case c: K '= K + 2, P' = 4000 + Pf for K '= O and K' = 1 Sample YN 'is positive for K' = 2 and K '= 3

Sample YN 'is negative.

  These findings lead us to code K as follows:

boasts:

K S Q

0 O O

1 O 1

2 1 O

3 1 1

  S represents the sign of the sample associated with the pair (K, P) namely S = O for a positive sample, S = 1 for a negative sample Q eb of K parity, defines the operations

giving the pair (K ', P').

  Figure 3 summarizes the flowchart to deduce each

  than triplet (P ', Q', S ') of the preceding triplet (P, Q, S).

  The tone generator according to the invention thus allows

  to generate any integer value of frequency used in tele-

  accurately address and generate any real value of

  frequency with an error of less than 0.5 Hz.

  The device of the invention applies to the generation of

  tones, inter-center signaling, testing of

riels etc

251229 J

Claims (3)

  I A method of coding a frequency, said frequency pre-   An integer value in the 1 4000 Hz L band, consisting in using the periodicity of the sine function on (0, 21 r), in matching each frequency f with samples, each sample being represented by a triplet (P, Q, S), P representing the rank, Q and S making it possible to locate the quadrant to which said sample belongs on (0, 2 W), characterized by storing -i samples   of the 1 Hz sinusoid on (0,), where T is the sampling period   tracing, where r is a natural integer divisor of l = 8000, and T is made to correspond to any sample YN representative of the frequency f (YN = sin 2 and P T), a triplet sample i   (P, Q, S) selected from said Ag samples, said correspon- dance being one-to-one.   2 Method of encoding a frequency according to claim 1   characterized in that said integer r is chosen equal to 1.   * 3 digital tone generator device implementing the method of claim 1 characterized in that it comprises means for storing samples of the 4 s LT sinusoid 1 Hz on (0, 1), r being a natural number, means 2 'for matching to each representative sample of each tone frequency two samples of rank P1 and P2, represented by the triplets (P1, Q1, -S1) and (P2, Q2, 52) among the 1 samples, said correspondences being one-to-one, T being the sampling period, means for characterizing the transmission state or blocked by another dynamic D parameter means for deriving from the triplet (P 1, Q 1, Si) (resp. 2, Q 2 '52) of   the previous state and the frequency fl (resp 2) forming the tonic   lity, the new triplet P'l, Q'I, S'1 (resp (P'2, Q'2, S'2))   characteristic of the frequency fl (resp f 2).